[sdiy] Harmonic bandwidth

[Tom Wiltshire]

On 9 Jan 2008, at 15:51, Scott Gravenhorst wrote:

> Nicholas Gregorich <nicksdsu at mac.com> wrote:
>> Scott Gravenhorst wrote:
>> One thing I just noticed, in the documentation for the sin()
>> function for C, it says "The sin() > function may lose accuracy
>> when its argument is far from 0.0".
>>
>> I've done the same test [32 harmonics with 0 phase versus 32  
>> harmonics
>> with constant random phases] using MATLAB and heard significant
>> differences between the 2 clips [which really surprises me]. I don't
>> think what you're hearing is inaccuracies from the C sin()
>> implementation. MATLAB uses 64-bit floating point by default.
>>
>> Nick.
>
> *That* is surprising.  I would hope that MATLAB's implementation is  
> about as
> accurate as it can get.

Yes, I'd have thought so too. All this talk of floating-point errors  
sounds pretty fanciful to me. Human ears aren't THAT good! You can  
push a calculation on a computer pretty hard before you even hit  
integer limit errors, and I don't think I've ever managed to really  
stretch floating-point to breaking point, much less double precision.

Not saying it can't happen, but it's not where I'd start looking.

T.